AI for real estate in Sri Lanka earns its place on four unglamorous tasks and stops at a fifth: qualifying portal and WhatsApp enquiries, booking the viewing and sending the reminder, the follow-up after the viewing, and keeping listings current, while price, negotiation and anything touching title, valuation or a contract stays with a named human agent. The market is why the timing matters, because the Central Bank of Sri Lanka’s Real Estate Market Analysis for the first quarter of 2026 records the price index for new condominiums in the Colombo District at 284.4, up 18.5 per cent year on year, while the condominium sales volume index fell 15.2 per cent over the same period.
Every US dollar figure below converts at LKR 328.75, the Central Bank of Sri Lanka’s indicative USD/LKR spot rate for 10 September 2026. This is commercial information, not legal, tax or valuation advice.
Two hundred enquiries, six viewings, one sale
Picture a four-person brokerage off Havelock Road, two agents, one doing photographs and portals, one doing everything else. It is a hypothetical firm, not a client.
They take a three-bedroom apartment in Colombo 5 on a six-week exclusive, listed on two portals, a Facebook page and the office WhatsApp. It draws a couple of hundred enquiries. Six people view it. One buys.
That ratio is the office’s own count, not a published statistic, and yours will differ. Where the hours go does not. Most enquiries are answered, chased and forgotten by someone also driving to Battaramulla, and the ones that decayed were the good ones that waited until Tuesday.
The Central Bank numbers explain why that hurts more in 2026 than in 2023. Asking prices in the Colombo District rose 31.9 per cent for land, 11.0 per cent for condominiums and 6.4 per cent for houses in the year to the first quarter, and the district took 65 per cent of condominium transactions. Dearer stock, fewer sales, same two people.
Qualifying without offending is where property enquiry automation earns its money
Four things decide whether an enquiry is worth a Saturday: budget, location tolerance, timing and financing. Sri Lankan buyers will answer all four, in our experience of writing these scripts, provided nobody asks them like a bank.
An agent is good at this because it is patient and identical every time. It acknowledges the listing, states the two or three facts a buyer asks next, puts one qualifying question, and waits. Open with the floor area and the maintenance charge, and ask about budget after the buyer has asked you something.
Three rules go into the specification. The agent may state only facts in the approved listing record. It may never say a price is negotiable, and never characterise value, yield or a comparable sale. And it hands over to a named person the moment a buyer asks about title, tax, the deed or whether an apartment is a good investment.
Owners ask here whether they need a chatbot. They do not, and the difference between a scripted tree and an agent that reads, decides and drafts is the subject of AI agents versus chatbots.
Booking the viewing, and the reminder that stops the empty Saturday
Booking is mechanical, and it is where the second largest block of time disappears. Offer two slots, hold one, put it in the agent’s calendar, send the location pin and the building’s access rules, then remind the evening before and two hours out. No-shows are the tax on doing this by memory, and a reminder naming the time, the gate and the person who will meet them is the cheapest thing in this article.
Most of it lands on WhatsApp, a channel with rules rather than a free broadcast list. Opt-in, the twenty-four hour service window and Meta’s per-message pricing are set out in WhatsApp automation in Sri Lanka, which you should read before anybody connects an automation to your office number.
The follow-up after the viewing is where deals are lost
Ask any agent what they know they should have done last month. It is the message two days after a viewing: you saw the Colombo 5 apartment on Saturday, is the parking the problem or the price, and shall I show you the two similar ones on Thursday.
Real estate lead follow up automation is worth more than the first reply, and almost nobody runs it, because it feels like chasing. It is also the only part of the pipeline where the buyer has already spent an hour on you. An agent that holds every viewing, drafts a follow-up on day two, a second on day seven and a check at day thirty, logs the reply and escalates by name, removes the queue rather than speeding it up. The buyer who said no in March buys in September, and only a list remembers that.
| Task | Channel | Agent role | Who approves | Hours a month, our estimate |
|---|---|---|---|---|
| First reply and qualifying | ikman, LankaPropertyWeb, WhatsApp, Facebook | Drafts facts from the listing record, asks budget, area, timing and financing one at a time | Agent on duty, for anything priced | 35 to 60 |
| Booking the viewing and the reminder | WhatsApp, calendar | Offers slots, confirms, sends the pin and access rules, reminds the night before | Template approved once | 12 to 20 |
| Post-viewing follow-up | WhatsApp, email | Drafts day two, seven and thirty messages, logs replies, escalates by name | Agent on duty, by exception | 15 to 25 |
| Listing updates and portal refresh | Website, portal accounts | Drafts descriptions, stages the update, flags stale listings | Principal, every price and new listing | 10 to 18 |
| Price, negotiation, title, valuation, contract terms | Any | Nothing: drafts nothing, forwards to a person | Named human agent, always | 0 |
Those hours are our estimates from mapping work, not measurements of your office. The first four rows come to 72 to 123 hours.
Listing updates, and what the web agent is actually for
The fourth task matters more to developers than to brokerages. A sales office with forty units has a website, a portal account or two and a price list that changes, and the gap between them is where trust goes. A buyer who finds a sold unit still advertised assumes the rest of the page is fiction.
A web agent takes the listing record as the single source of truth, drafts the description, stages the change and waits for approval before anything goes live. It also notices the listing untouched for eleven weeks and the unit marked available that your reservation sheet says is gone. That is a page treated as a working system rather than a leaflet, the case we made in your website was never a brochure.
The listing text is yours in law, not the portal’s. ikman’s terms and conditions state that “Advertisers and users are responsible for ensuring that advertising content, text, images, graphics, video (“Content”) uploaded for inclusion on ikman complies with all applicable laws”. A generated description that overstates an extent or invents an approval is your problem, so the principal approves each one.
Where the human agent decides, and why AI in real estate stops there
specshop.dev is an AI consultancy and AI agency in Colombo, Sri Lanka, led by Janaka Ediriweera, Principal AI and Product Management Consultant. In property specifications we write one line above all the others: an agent may describe a property and may never characterise a transaction.
That covers price and any hint of negotiability, valuation and yield, comparables, title and extent, the deed, encumbrances, planning approvals, taxes and duties, tenancy terms and every clause of a sale agreement. It also covers a regulated act. Sri Lanka’s Financial Intelligence Unit sets out that real estate agents are designated non-finance businesses under section 33(h) of the Financial Transactions Reporting Act No. 06 of 2006, must appoint a senior compliance officer, must carry out customer due diligence under the 2018 rules, and must report a suspicious transaction within two working days. An agent may collect enquiry details that a person then verifies. It does not perform due diligence, judge what is suspicious, or sign anything.
No licensing examination stands behind the ordinary Sri Lankan estate agent, which cuts the opposite way from what people assume. Where the regulator is silent, the discipline comes from your own written rules.
What it costs, in rupees
A support and sales agent, the roster post that fits an enquiry desk, is US$1,225 a month, LKR 402,700, with a one-off recruitment fee of US$9,500, LKR 3,123,125, covering deployment, connection to your WhatsApp number, inbox and calendar, a 30-day probation with weekly scorecards, and the first month’s pay. A web agent is the same salary. If the approval rate has not improved on its day-one baseline by day 30, you do not confirm the hire and the balance is never invoiced. The terms sit on the page where you hire an AI agent.
Now the sum. According to the Department of Census and Statistics’ Labour Force Survey Annual Report 2024, the mean monthly gross salary in services was LKR 62,176. Loaded with EPF at a minimum of 12 per cent and ETF at 3 per cent, that is LKR 71,502 a month, about LKR 409 an hour over a 175-hour month. LKR 402,700 divided by LKR 409 is roughly 985 staff hours, and the table tops out near 123, so this never pays as wage arbitrage.
It pays on one thing: the enquiry answered at nine on a Sunday evening, while the buyer is still deciding which of four agents to reply to. Test it against your commission on one extra sale a quarter before anybody quotes you a saving.
What to do this week
Count before you shop. For ten working days log every enquiry: channel, hour, listing, what it asked, who answered, how long it took, and whether it ever got a second message. The second message column will embarrass you, and it is also your business case.
Then write one page of rules for the first reply alone: which facts may be stated, which words may never appear, three worked examples, and who approves a price. Take that page to a provider before you take a demo. If you would rather have the counting done, our workflow mapping sprint is US$2,300, LKR 756,125, for two video sessions and a written report in 48 hours, credited against the recruitment fee if you hire within 30 days, and the first thirty-minute call is free. We do this work as an AI agency in Colombo, and the firms that get value from it arrive with a count.
Questions people actually ask.
Can AI answer property enquiries in Sri Lanka?
Yes, and the first reply is the sensible place to start, because the questions repeat and the value is in the answer landing while the buyer is still comparing listings. Keep the agent to facts that appear in your approved listing record, have the agent on duty approve anything carrying a price, and route title, tax and valuation questions to a named person immediately.
How do real estate agents in Sri Lanka use AI?
In practice the useful work is unglamorous: drafting first replies and qualifying questions on portals and WhatsApp, offering and confirming viewing slots with reminders, following up two days and seven days after a viewing, and keeping listing text and availability current across the website and portal accounts. Everything about price, negotiation, title, valuation or a contract term stays with the human agent whose name is on the deal. The level of ambition is not exotic anywhere: the National Association of Realtors’ 2025 Realtors Technology Survey, published on 18 September 2025, found 20 per cent of members using AI daily, 22 per cent weekly and 27 per cent a few times a month, with 46 per cent using AI-generated content such as listing descriptions.
Can AI qualify property leads on WhatsApp?
Yes, and it is the channel where most Sri Lankan property enquiries actually arrive, so it is usually the first one connected. Ask one qualifying question at a time rather than a form, respect Meta’s opt-in requirement and the twenty-four hour service window, and never let an automated message be the last word on a price or a negotiation.
Will AI replace real estate agents?
No, because the parts of the job that carry money and risk are exactly the parts an agent must not touch: price and negotiation, valuation, title and extent, and every clause of a sale agreement. What it replaces is the retyping, the forgotten follow-up and the stale listing, which is where a small brokerage currently loses most of its week.
How much does real estate automation cost in Sri Lanka?
A support and sales agent covering qualifying, viewings and follow-up is US$1,225 a month, LKR 402,700, with a one-off recruitment fee of US$9,500, LKR 3,123,125, that includes deployment, a 30-day measured probation and the first month’s salary. At the Department of Census and Statistics’ 2024 services wage, loaded to about LKR 409 an hour, that monthly figure is roughly 985 staff hours, so the case is made on enquiries answered and sales completed rather than on headcount removed.
Sources
- Real Estate Market Analysis, Condominium Market Survey First Quarter 2026, Central Bank of Sri Lanka — “The Price Index for New Condominiums in the Colombo District reached 284.4 in 2026 Q1, reflecting a year-on-year increase of 18.5 per cent”; the Condominium Property Sales Volume Index “recorded a year-on-year decrease of 15.2 per cent”; the Colombo District accounted for 65 per cent of condominium transactions; asking price indices for lands, houses and condominiums in the Colombo District rose 31.9, 6.4 and 11.0 per cent year on year; 20 developers took part in the survey. Fetched 11 September 2026.
- Real Estate, Designated Non-Finance Businesses and Professions, Financial Intelligence Unit of Sri Lanka — real estate agents are designated under section 33(h) of the Financial Transactions Reporting Act No. 06 of 2006; Rule 41 of the Designated Non-Finance Business (Customer Due Diligence) Rules No. 1 of 2018 requires appointing “a senior management level officer as the CO to deal with AML/CFT related matters”; customer due diligence data points and enhanced due diligence for high-risk customers; suspicious transaction reports to be made “as soon as practicable…but no later than two working days” under section 7 of the FTRA. Fetched 11 September 2026.
- Terms and Conditions, ikman — “Advertisers and users are responsible for ensuring that advertising content, text, images, graphics, video (“Content”) uploaded for inclusion on ikman complies with all applicable laws”; user email addresses are not publicly displayed and messages are passed through ikman. Fetched 11 September 2026.
- Realtors Embrace AI, Digital Tools to Enhance Client Service, National Association of Realtors, 18 September 2025 — from the 2025 Realtors Technology Survey, “20% use them daily, 22% weekly, 27% a few times a month, and 32% have not yet used AI”, and “46% report using AI-generated content (e.g., for listing descriptions)”. Cited as United States data for context only. Fetched 11 September 2026.
- Sri Lanka Labour Force Survey Annual Report 2024, Department of Census and Statistics — mean monthly gross salary of monthly wage or salary earners LKR 58,292, services sector LKR 62,176. Fetched 10 September 2026.
- Employees’ Provident Fund, Central Bank of Sri Lanka — minimum contribution of 12% from the employer, used to load the hourly wage. Fetched 10 September 2026.
- Employer Details, Employees’ Trust Fund Board of Sri Lanka — employer contribution of 3% of total monthly earnings, used to load the hourly wage. Fetched 10 September 2026.
- Daily Indicative USD/LKR Spot Exchange Rates, Central Bank of Sri Lanka, 2026 — the indicative rate of LKR 328.75 per US$1 on 10 September 2026, used for every conversion in this post. Fetched 10 September 2026.